The question comes to us every week: «How do I know if ChatGPT recommends me?»
The short answer: you can know today, in 30 minutes and at no cost.
The long answer: measuring it once is almost useless. AI visibility changes week by week, and what matters is the trend.
Let’s break it down.
How to see a brand’s visibility in ChatGPT?
The manual method in 5 steps
Step 1: Define the prompts your client would use
Don’t ask «Do you know [tu marca]?». That measures fame, not business. Ask what a buyer who doesn’t know you yet would ask:
- «What [categoría de producto] do you recommend for [caso de uso]?»
- «Best [categoría] in [país] in 2026»
- «[Competidor] vs alternatives, which one should I choose?»
- «I need [resolver problema X], which company or tool should I use?»
Write between 10 and 20 prompts. If you serve multiple countries, duplicate the important ones by market.
This list is your asset: these are the same prompts that you will monitor every week. Without knowing how your market asks, everything else is guesswork.
Step 2: Ask in clean sessions
ChatGPT personalizes based on your history. If you’ve talked about your brand before, the result comes out contaminated.
Use a temporary chat (or an account without history), enable web search, and launch each prompt in a new conversation.
Since the launch of the integrated ChatGPT search, a part of each response comes from the web in real time, so the cited sources matter as much as the answer.
Step 3: Record 4 things from each response
Set up a spreadsheet with one row per prompt and these columns:
| Column | What to note |
|---|---|
| Do you appear? | Yes / No |
| Mention position | 1st, 2nd, 3rd mark… or «at the end» |
| Competitors mentioned | All, in order |
| Cited sources | Domains of the response links |
| Sentiment | How it describes you: positive, neutral, with caveats |
This is how it looks in practice:
The sources column is gold: those domains are your list of sites to get mentions. If three responses cite the same list of «best tools», you already know what your next target is. We tell you how to work on it in our guide on how to appear in ChatGPT.
Step 4: Repeat each prompt 3 times
ChatGPT does not respond the same way twice. A brand may appear in 2 out of 3 attempts of the same prompt.
Launch each prompt at least 3 times in separate sessions and calculate your visibility as a percentage: if you appear in 18 out of 30 responses, your visibility is 60% in that set of prompts.
Step 5: Repeat the exercise every week
An isolated photo says nothing: you might have a good week or a bad one.
The useful signal is the trend over 4 to 8 weeks, especially after publishing content, getting a mention, or seeing a model update. Schedule 30 minutes on the same day each week.
Where the manual method falls short
The method above works and is the best way to start. But it has four limits that appear quickly:
1. Variability forces you to do many repetitions. With 3 attempts per prompt, you have an approximation; for a stable figure, you need many more executions than anyone does manually.
2. The country changes the response. A user in Mexico and another in Spain receive different brands with the same prompt. Measuring a single market manually is already costly; measuring 5 is unfeasible.
3. ChatGPT is not the only one that matters. Your client also asks Gemini, Perplexity, and Google AI Overview, and each cites different sources and brands.
4. It doesn’t scale over time. 20 prompts x 3 repetitions x 4 models x each week means hundreds of manual queries per month. No one can sustain it for more than three weeks, and the point was precisely in the consistency.
How to automate it: Crecerank’s visibility analysis
Crecerank does exactly the manual process above, every day and without you having to remember:
- Visibility %: launches your prompts daily in ChatGPT, Gemini, Perplexity, and Google AI Overview and gives you your appearance percentage with temporal evolution, by model and by country (11 countries in Latin America and Spain).
- Competition: automatically detects which brands are mentioned alongside yours and your share of voice against each one.
- Source analysis: lists the domains that the AI cites when responding, ordered by frequency. It’s your list of mention targets made automatically.
- Actionable GEO: turns all of the above into a prioritized plan: what content to create, which sources to target, and what to correct first.
The Starter plan costs US$49 per month with 1 brand and 10 prompts, and you can try it for 14 days free without a card.
Our honest recommendation: first do a manual round with the method above. If with that first snapshot you see that the topic matters for your business (it almost always happens one of two ways: you don’t appear where your competitor does, or you are described incorrectly), automate.
Conclusion
Analyzing your visibility in ChatGPT does not require a budget: it requires a method.
Real client prompts, clean sessions, systematic recording, repetitions, and weekly consistency. With that, you already know more than 95% of your competition.
The limit is not the technique, it’s the time. When the weekly exercise starts to fall apart, that’s the moment to automate.
Do you want your first serious measurement today? Crecerank runs your prompts on 4 AI models and shows you in minutes where you appear, where you don’t, and what sources you need to work on. 14 days free, no credit card.
